User Preference based Cross Domain Recommender System

Abstract

Recommender systems have developed with the advent of technology. With the increasing amount of data on web, it has become difficult to provide quick and user satisfactory recommendations. Inspite of the recommendation systems having achieved tremendous success in various domains, however improvement is still required in cross-domain recommendation field. On account of the increase in volume of music data stored online, opportunities have opened up to implement music recommender systems among users. It is always a difficult task to recommend appropriate music to the users. It becomes easier to recommend music if certain context based information is provided. This leads to cross domain recommendation system.

Cross Domain Recommendation System recommends two different items from two different domains. This dissertation focuses basically on cross domain recommendation comprising of selecting two things at a time from two separate domains and recommending them together. The system will suggest music tracks to the user on the basis of the place along with consideration of user's preference of music tracks. Hence, the combination of Place-Music pair is recommended by using different approaches using the tags attached to music tracks and places by the users.